Effect of radio reflections / multi path on received message

[Robert]

Hi

We use the NRF51 in a mesh sensor network. We noticed in a trailer suited for cold chain transport (double metal walls with thermal isolation between) that the radio has problems with receiving message. Even at short distances (< 5 meter) we missed a lot of message at 2Mbit/s mode (90% of send message either missed or CRC erros). This only occurs in the trailer. The RSSI of the received message was around -40 dBM, so in general it is not a distance problem. We expect some effect of reflected message. However, distance (compared to speed of light) are to short to expect pathlengt delay's in the order of one bit time. Reducing the data rate to 1Mbit/s improves (less missed messages), but still yield very variable results. However at 250kbit/s we had 99 to 100% succes rate, no missed message nor CRC erros. The RSSI receive strenght of the message is still around -40 dBM.

We don't understand the effect we see completly. Can somebody explain what is happening with the FSK modulation and relfections and why message are received at 250 kbit/s oke ? Is there a way to achieve the same performance at 2Mbit/s by a certain radio setting ?

[Robert]

Hello Øyvind

Thanks for your feedback. Some remarks and additional questions.

We are aware of the possible interfernce issue due to multipath. That is the main reason why we selected a mesh solution. The mesh solution should make the data distribution independent from the actual position of a single transmitter and a single receiver.

Therefore we did all the measurement initially at the same position. In the same position we did the tests at 2Mbit/s, 1Mbits/s and 250 kbits/s, with the results reported above. Thereafter we moved the setup several times in several directions with half a wavelenght, with similair results. This gave us the idea that we are not having peak or interference at the 2.4 Ghz wave, i.e that the demodulator could follow the 2.4 Ghz wave. (Is there a way to see if the NRF radio has difficulties to lock the demodulator to the 2.4 Ghz wave ?)

As stated, the improvement from 2Mbit/s to 250 kbit/s where done at exactly the same postion. This is imho not in line with the remark that 3MHz coherence bandwidht could be expected. The difference between 2Mbit/s and 250 kbit/s at the same position gave us the idea that "bit" demodulation errors occur. For a pathlenght difference of 15 meter (extra lenght when traveled once in the trailer ) the reflected wave would have a delay of 50 nano seconds. This is 10% of period at 2Mbit/s and only 1.25% at 250 kbit/s. Could this be the case ? If so, is there a way to make the 2Mbit/s demodulation in the radio less sensitive to this effect (by filtering in the antenne circuit ?)

[Øyvind Karlsen]

Hi,

I agree that it is not in line with the 3 MHz statement, on closer inspection the measurements that state this are from an open office, which is very different from your scenario.

There are expensive ways to circumvent this, however it is likely better to cope with the possibility of reduced datarates, ie. switch to a lower datarate when you get large packet drops. The problem you are facing is related to a physical limitation that causes symbol interference. Traditionally Inter Symbol Interference (ISI) is addressed by using an equalizer or applying matched filtering. Finally you can attempt to obtain some antenna diversity.

Øyvind